Re: b43 with upstream-davem stacktrace

[Michael Buesch <mb@bu3sch.de>]

On Monday 15 October 2007 11:05:26 Richard Purdie wrote:
> On Sun, 2007-10-14 at 11:42 +0200, Johannes Berg wrote:
> > On Sat, 2007-10-13 at 17:26 +0200, Michael Buesch wrote:
> > 
> > > I also saw this sometimes, but I have absolutely no idea how that happens.
> > > Someone any idea?
> > 
> > Nope, but adding more CCs :) I briefly looked at the code and couldn't
> > track it. Got a powerpc oops with disassembly?
> 
> It appears to be having a problem in the strcmp in
> led_trigger_set_default() which is odd. Code above that checks if one of
> the options is NULL so all I can think it either the default_trigger
> string being passed in is invalid or there is a trigger with an invalid
> name registered with the system.
> 
> Is this an SMP system and would something else be registering an LED
> trigger on another processor or similar? There is locking there which
> should be sufficient though. So no, I'm not sure whats going on
> either...
> 

I think this is some kind of a race, as it does only happen rarely and
is not really reproducible.


-- 
Greetings Michael.